[Esip-preserve] A Minor Note on Nomenclature

[Curt Tilmes]

On 08/24/2010 11:40 AM, alicebarkstrom at verizon.net wrote:
>From a personal perspective, I'd prefer to keep the term "provenance"
>(or - more precisely - "production history provenance") devoted just
>to files and source code PGEs that directly enter data production.
>In the OAIS RM, the other documents, such as calibration data,
>calibration plans and reports, Point Spread Function data and
>reports, and so on, seem to me to be part of the Representation
>Network that's needed for understanding data.  However, unless those
>documents are part of the production process, it seems to me to be
>inappropriate to include them in "provenance tracking".
>
>If need be, I can see what OAIS RM has to say.
>
>Bruce B.

Well, as the "community", we can establish consensus to use whatever
nomenclature we want.  If we want to define it like that, fine.  I
prefer using "lineage" or "production history" for the limited term,
and grouping all of the other stuff (calibration, analysis papers,
post-processing validation, etc.) into "provenance".

If you like the term "provenance" to be equal to "lineage", what do
you use for all the other stuff?  Do you really think we will have an
easier time selling these ideas if we call it the "Representation
Network"?

I think an "audit" history is an important part of provenance as well.

Lineage/Production History: I fed X,Y to process A, version 1.7 and
got file Z.

Audit: I got file X from organization Q on Jan 7, 2010 at 4:50pm.  I
compiled A on host H with compiler C version 7.2.  I started the
process on Jan 17 at 2:25pm on host X.  It ran for 27 minutes and used
50 MB of memory.

Validation: I compared file Z with ground station G and produced a
paper P describing the results.

Provenance: All of the above.


I liken it to the historical art world.  Much of the value of the art
isn't in the painting itself, or even in the production history (the
paint, canvas and paint brushes), but in the history of ownership
(Fred bought the painting in 1752 from the artist, his family donated
it to a museum in 1780, etc.) and even more in the analysis and
appraisals (the brush strokes match Rembrandt's style, the chemical
analysis of the paint and canvas match what we we would expect, etc.)
All of that related information that contributes to our trust and
ultimately, the value of the art I would group into "provenance".


I see your point -- we are stretching the traditional meaning of the
word, but so what?  Let's stretch it into something useful for our
domain.

Curt